U.S. patent application number 09/902577 was filed with the patent office on 2003-01-16 for system and method for managing transportation demand and capacity.
Invention is credited to Clarke, Lloyd, Gamble, A. Bruce, Jancik-Bailey, Jennifer, Orchard, Ryan.
Application Number | 20030014288 09/902577 |
Document ID | / |
Family ID | 25416049 |
Filed Date | 2003-01-16 |
United States Patent
Application |
20030014288 |
Kind Code |
A1 |
Clarke, Lloyd ; et
al. |
January 16, 2003 |
System and method for managing transportation demand and
capacity
Abstract
The present invention comprises a system and method for managing
transportation demand and capacity. The present invention allows a
carrier to perform rapid and accurate determinations of the
profitability of accepting various load transporting opportunities
while taking into account the effect of taking a particular load on
the entire carrier network. The present invention makes these
profitability determinations based on a network model which is
continually updated to account for changing market conditions and
the effects of real-time events. The method of the present
invention includes creating a dynamic network flow model comprised
of multiple nodes, each node representing a specific location at
specific time. The marginal value of a unit of capacity (e.g. a
truck, a trailer, a rail car, a plane, etc.) at each node is
calculated by solving the dual of a linear program associated with
the network flow model. A matrix is created by a dynamic network
value engine (NVE). The matrix contains a marginal value for a unit
of capacity for each node in the network flow model up to some
predetermined time in the future. The profitability of transporting
a given load from a source node to a destination node is made based
on the revenue minus the cost plus the marginal value of a unit of
capacity at the destination node minus the marginal value of a unit
of capacity at the source node. The marginal value of a unit of
capacity at a given node is obtained from the matrix. The dynamic
NVE periodically and continually updates the matrix to account for
changing market conditions. Transportation decisions are then made
based upon the profitability determinations. The present invention
also includes a "webcrawler" feature. The webcrawler searches a
database offers by shippers to have loads shipped. The webcrawler
determines the profitability of each offer and prioritizes the
offers based on profitability.
Inventors: |
Clarke, Lloyd; (Green Bay,
WI) ; Gamble, A. Bruce; (Green Bay, WI) ;
Jancik-Bailey, Jennifer; (De Pere, WI) ; Orchard,
Ryan; (Edmonton, CA) |
Correspondence
Address: |
SWIDLER BERLIN SHEREFF FRIEDMAN, LLP
3000 K STREET, NW
BOX IP
WASHINGTON
DC
20007
US
|
Family ID: |
25416049 |
Appl. No.: |
09/902577 |
Filed: |
July 12, 2001 |
Current U.S.
Class: |
705/7.26 ;
705/400; 705/7.31; 705/7.37 |
Current CPC
Class: |
G06Q 10/06316 20130101;
G06Q 10/06375 20130101; G06Q 30/0202 20130101; G06Q 10/08 20130101;
G06Q 30/0283 20130101; G06Q 10/047 20130101 |
Class at
Publication: |
705/7 ;
705/400 |
International
Class: |
G06F 017/00; G06F
017/60 |
Claims
1. A method of managing transportation demand and capacity,
comprising: creating a network flow model comprised of a plurality
of nodes, each node representing a specific location; calculating
the duals of a dual linear program based on the network flow model
to determine the marginal value of a unit of capacity at a source
node and the marginal value of a unit of capacity at a destination
node; calculating the value of transporting a load from the source
node to a destination node based on the marginal values of a unit
of capacity at the source node and destination node; and making a
transportation decision based on the calculated value of
transporting the load.
2. The method of claim 1, wherein the network flow model is
comprised of a plurality of nodes, each representing a specific
location at a specific time.
3. The method of claim 2, wherein the source node and the
destination node are connected by an arc, the arc having a variable
associated with the arc, the variable representing a number of
units of capacity to be moved between the source node and the
destination node.
4. The method of claim 3, wherein the network flow model includes
constraints at each node representing conservation of flow.
5. The method of claim 4, wherein the arc has an upper bound
representing the demand for loads to be transported between the
source node and the destination node, and the arc has a lower bound
representing commitments for loads to be transported between the
source node and the destination node.
6. The method of claim 5, further comprising: forecasting the
demand between the source node and the destination node based on
historical data.
7. The method of claim 5, wherein the arc has an associated average
revenue and average cost.
8. The method of claim 2, further comprising: solving the dual of a
linear program associated with the network flow model to determine
the marginal value of a unit of capacity at the source node and the
marginal value of a unit of capacity at the destination node.
9. The method of claim 2, wherein the source node and the
destination node are connected by a plurality of arcs, each arc
having an associated revenue and an associated cost.
10. The method of claim 2, further comprising: creating a matrix
containing the marginal value of a unit of capacity at each node in
the network flow model up to a predetermined time in the
future.
11. The method of claim 10, further comprising: periodically
updating the matrix values by resolving the duals of a linear
program associated with the network flow model.
12. The method of claim 11, further comprising: calculating the
profitability of transporting a given load from A to B according to
the equation: profitability=Revenue-Cost-Val (A)+Val (B), wherein
Val(A) and Val(B) are the marginal value of a unit of capacity at
location A and location B, respectively.
13. The method of claim 12, wherein the marginal values of a unit
of capacity are obtained from the matrix.
14. The method of claim 13, further comprising: using the
profitability calculation to make at least one of the following
transportation decisions: a) deciding whether or not to accept an
offer to transport a load at a specified contracted price over a
specified time period; b) prioritizing a plurality of offers to
transport loads based on profitability; c) determining a contracted
price to offer for transporting a load; d) determining a price to
offer a shipper, for soliciting the shipper to transport a load by
an idle unit of capacity; e) determining a spot price for
transporting a load; f) selecting a mode of one of solo, team,
rail, third party, regional or Canadian; and g) assigning a
specific unit of capacity and a specific driver to a particular
load.
15. The method of claim 13, further comprising: searching a
database containing a plurality of offers to have loads shipped;
determining the profitability of each offer; and prioritizing the
offers based on profitability.
16. The method of claim 15, further comprising: spidering a
database connected to a network to search for offers.
17. The method of claim 1, wherein the transportation decision is
used in a scenario evaluator.
18. A method of managing transportation demand and capacity,
comprising: creating a matrix containing the marginal value of a
unit of capacity at each node in a network flow model up to a
predetermined time in the future by solving the duals of a linear
program associated with the network flow model; periodically
updating the marginal values in the matrix by resolving the duals
of a linear program associated with the network flow model;
calculating the profitability of transporting a load based on the
marginal value of a unit of a capacity at a source node and the
marginal value of a unit of capacity at a destination node; and
making a transportation decision based on the profitability
calculation.
19. A method of managing transportation demand and capacity,
comprising: calculating the marginal value of a unit of capacity at
a source node and a destination node in a network flow model by
solving the duals of a linear program associated with the network
flow model; calculating the profitability of transporting a load
based on the marginal value of a unit of a capacity at the source
node and the marginal value of a unit of capacity at the
destination node; and making a transportation decision based on the
profitability calculation.
Description
BACKGROUND OF THE INVENTION
[0001] The truckload industry is highly competitive with very thin
profit margins. In such an industry, it is critically important
that a company fully understand the potential profitability (or
lack thereof) of every piece of business that comes its way.
Current methods for evaluating profitability are cumbersome and
don't accurately account for real-time events and current market
conditions. What is needed is a tool that allows a truckload
company or any other type of carrier to quickly and accurately
calculate an estimated profit of a given load, taking into account
all effects on the carrier's surrounding network.
SUMMARY OF THE INVENTION
[0002] The present invention comprises a system and method for
managing transportation demand and capacity. The present invention
allows a carrier to perform rapid and accurate determinations of
the profitability of accepting various load transporting
opportunities while taking into account the effect of taking a
particular load on the entire carrier network. The present
invention makes these profitability determinations based on a
network model which is continually updated to account for changing
market conditions and the effects of real-time events.
[0003] The method of the present invention includes creating a
dynamic network flow model comprised of multiple nodes, each node
representing a specific location at specific time. The marginal
value of a unit of capacity (e.g. a truck, a trailer, a rail car, a
plane, etc.) at each node is calculated by solving the dual of a
linear program associated with the network flow model. A matrix is
created by a dynamic network value engine (NVE). The matrix
contains a marginal value for a unit of capacity for each node in
the network flow model up to some predetermined time in the future.
The profitability of transporting a given load from a source node
to a destination node is made based on the revenue minus the cost
plus the marginal value of a unit of capacity at the destination
node minus the marginal value of a unit of capacity at the source
node. The marginal value of a unit of capacity at a given node is
obtained from the matrix. The dynamic NVE periodically and
continually updates the matrix to account for changing market
conditions. Transportation decisions are then made based upon the
profitability determinations.
[0004] Each node in the network is connected by an arc. The arc has
an associated variable representing the number of units of capacity
to be moved between the connected nodes. The network flow model
includes constraints at each node representing conservation of
flow.
[0005] Each arc has an upper bound representing the demand for
loads to be transported between the source node and the destination
node, and the arc has a lower bound representing commitments for
loads to be transported between the source node and the destination
node. The demand for loads to be transported on a particular arc is
determined by demand forecasting based on historical data. Each arc
has an associated average revenue and average cost. Two nodes can
also be connected by multiple arcs, each arc having an associated
revenue and an associated cost.
[0006] A holistic model can also be used. The holistic network flow
model does not have different nodes for different times, but only
includes a single node for each location. The holistic model is
simpler because it doesn't take into account changing conditions
over time. The holistic model provides a simpler, less
computationally intensive model that can be useful as a scenario
evaluator for decision making on contracted pricing and other
long-term decision making. Both the holistic and dynamic models can
be used not only for making current transportation decisions, but
they also can double as scenario testers to help answer a range of
longer-term tactical and strategic questions.
[0007] The profitability determinations can be used to make a
variety of load transportation decisions such as a) deciding
whether or not to accept an offer to transport a load at a
specified contracted price over a specified time period; b)
prioritizing a plurality of offers to transport loads based on
profitability; c) determining a contracted price to offer for
transporting a load; d) determining a price to offer a shipper, for
soliciting the shipper to transport a load by an idle unit of
capacity; e) determining a spot price for transporting a load; f)
selecting a mode of one of solo, team, rail, third party, regional
or Canadian; and g) assigning a specific unit of capacity and a
specific driver to a particular load.
[0008] The present invention also includes a "webcrawler" feature.
The webcrawler searches a database offers by shippers to have loads
shipped. The webcrawler determines the profitability of each offer
and prioritizes the offers based on profitability. The webcrawler
can also spider into business-to-business databases connected to
the world wide web, and find the best offers.
BRIEF DESCRIPTION OF THE DRAWINGS
[0009] FIG. 1 illustrates a block diagram depicting the network
value engine and order process of the present invention.
DETAILED DESCRIPTION OF THE INVENTION
[0010] The following terminology will be used herein. A "carrier"
is a company in the business of transporting goods. For example,
the carrier may own trucks, trailers, rail cars, airplanes, and so
forth. A "shipper" is a customer who has a load that needs to be
shipped and wants the carrier to ship the load from point A to
point B. For example, the shipper could be a large retailer or
chain of supermarkets. Although many examples will be given below
with regards to the trucking industry, the present invention is not
limited to the truckload industry but can be used by any carrier in
the business of transporting goods, people, or any other item.
Lastly, a "lane" refers to a particular shipment route, for
example, Denver-to-Los Angeles represents one particular
"lane."
[0011] The standard definition of profit is simply total revenue
minus total cost. In the truckload industry, however, there is an
additional complication. The complication is the fact that moving
freight from Point A to Point B also moves a unit of capacity (e.g.
a truck, a rail car, a trailer, a plane etc.) from Point A to Point
B. Thus, when evaluating the profitability of moving a load from
Atlanta to Baltimore, not only must the direct revenues and costs
be considered, but also the difference in value of having an extra
truck in Baltimore vs. having an extra truck in Atlanta. Therefore,
to determine the profitability of moving a load from A to B, the
following equation must be used:
Profitability of a load AB from A to B=Revenue (AB)-Cost (AB)-Val
(A)+Val (B)
[0012] where Val(i) is an estimate of the marginal value of having
a truck at location i.
[0013] In order for the carrier to make wise decision making
regarding which loads to ship and how much to charge, the carrier
needs to have a fast and accurate method of determining the
profitability of transporting a particular load according to the
above equation. Determining direct revenue and cost is relatively
straightforward. Therefore, determining the value of a load can be
reduced to the problem of determining the marginal value of a unit
of capacity in a given location, Val(i). The truckload industry
today uses a crude static model to determine the marginal value of
a unit of capacity. The static model, described below, evaluates
the value of the load in isolation from its impact on the overall
network. After the static model is described, two improved models
for determining Val(i) will be described, the dynamic and holistic
models, which provide a better determination of marginal value
because they take into accept a load's effect on the entire
network.
[0014] 1. Static Model
[0015] A first way of determining the marginal value of a unit of
capacity is a simple static model. The static model allows "static"
marginal values Val (A) and Val (B) to be calculated. With the
static model, each load is evaluated in isolation from its impact
on the overall network. Static marginal values Val(A) and Val(B)
are estimated by looking historically at the amount of profit
generated by the average truck after arriving at location B.
[0016] For example, the static model evaluates the historical data
showing the profits earned by trucks in the past when leaving a
location B. The static model determines the average amount of
profit generated by a truck, after leaving a location B over the
past month or the past quarter (or any desired time period). For
example, one data point might show that one truck departed from B
carrying a load to B' and obtained a profit of X. A second truck
carried a load from B to B" and earned a profit of Y. All of these
profits can be averaged over a desired time period to determine an
average profitability of a truck at location B; this in turn
determines Val(B), the marginal value of a truck at location B.
Alternatively, the marginal value Val (B) could be assigned to be
equal to the average profit of the next 2 loads of a truck after
leaving location B, or the next 3 loads, or any desired number of
loads after leaving location B.
[0017] The static model has the advantage that the marginal values
are easy to calculate if there is existing historical data (no
optimization algorithms are required). The disadvantage of the
static model is that it is a relatively crude method for
calculating marginal values Val(A) and Val(B), and does not take
into account current market conditions or the load's impact on the
entire network.
[0018] 2. Dynamic and Holistic Models
[0019] The dynamic model uses linear programming to evaluate a
load's impact on the entire network. The dynamic model is a
time-phased network flow model. The network flow model is comprised
of nodes, each node representing a particular location at a
particular time. As an example of a network flow model, suppose one
node represents the city of Denver at 9:00 am on Thursday, Jul. 5,
2001. There is also a node representing Chicago at the same date
and time, another node at Denver, another node at Los Angeles, and
so forth. At 1:00 pm on the same day, there could be a new set of
nodes for each of these locations. A truck leaving New York City at
9:00 am and arriving in Washington D.C. at 1:00 pm would be
traveling from a source node representing Washington D.C. at 9:00
am and arriving at a destination node representing New York City at
1:00 pm. The time spacing between nodes is a matter of design
choice. For example, the nodes could be separated by 4 hours, by 12
hours, by 1 day, by 1 month, by 1 year, etc.
[0020] All of the nodes are connected by arcs. Each arc represents
a feasible lane from an origin node at a given time, to a
destination node at another given time. Each arc has a variable
associated with the arc. The variable for the arc represents flow,
the number of trucks to be moved on that lane during that time
period. The linear program includes constraints at each node. The
constraints represent conservation of flow. The flow into each node
must equal the flow out. Arc upper bounds represent the total
available freight (i.e. the total demand for loads to be shipped)
for that arc. The arc lower bounds represent commitments; i.e.
freight that the carrier has already agreed to ship. Each arc has
an associated average revenue and average cost. By solving the
linear program, the optimal values of the variables can be
determined. The optimal values of the variables represent the
number of loads to ship for each arc to maximize profits, subject
to the constraints.
[0021] Associated with any linear program is another linear
program, called the dual. When taking the dual of a given linear
program, the given linear program is referred to as the primal. If
the primal is a max problem, the dual will be a min problem, and
vice versa.
[0022] The dual of the primal linear program described above is
determined. Each dual variable corresponds to a constraint in the
primal linear program. The dual variables represent the value of
adding capacity to that constraint. In other words, by solving the
dual LP, the marginal value of a unit of capacity at a specific
location at a specific time is determined.
[0023] As mentioned, the upper bounds on each arc represent the
total demand. These demand values must be determined by forecasts
of the demand. The demand for a particular lane at a particular
time is forecasted taking into account historical data, holiday
effects, and other known patterns or variation in demand over time.
For example, automobile manufacturers in Michigan have a shut down
every year in the summertime when they do the model year
changeover. All of the lines shut down for a time, so suddenly
there's very little demand. Factors such as this shutdown should
thus be accounted for in the demand forecast.
[0024] Each arc in the network flow model can potentially be split
into multiple arcs representing different revenue and cost buckets.
For example, suppose that there is one arc going from Denver to LA,
and a carrier has committed to move at least ten loads. There is a
total demand of 20 loads forecasted. Therefore, the carrier can
move between ten and twenty loads on that arc. Now, if only a
single arc is used, then the goal is to maximize the average
revenue minus cost on that arc. However, if the carrier has
different customers with very different revenues, the model could
include two different arcs representing two different customers,
with upper and lower bounds on each of those arcs. Suppose that the
carrier is committed to transport five loads for each customer and
might get up to as many as ten loads for each customer. However,
one customer provides $1.00/mile and the other customer provides
$1.30 per mile. The model could include two different arcs with a
lower bound of five and a upper bound of 10 on each of arc, a value
of $1.00 on one arc and a value of a $1.30 on the other arc. In a
similar manner arcs can be added to represent the possibility of
deadheading (moving an empty truck) from A to B, typically at a
negative profit.
[0025] The advantage of this dynamic model is that it incorporates
changing market conditions when calculating values. That is, a
given load may be undesirable today, but very desirable tomorrow
due to changes in market demand and capacity. The dynamic model is
valuable for day-to-day, operational decision making regarding
issues such as spot pricing, load acceptance, and other
applications described further below.
[0026] A simpler network flow model can also be used that does not
have different nodes for different times, but only includes a
single node for each location. This is called the "holistic" model.
The holistic model is simpler because it doesn't take into account
changing conditions over time. The holistic model provides a
simpler, less computationally intensive model that can be useful as
a scenario evaluator for decision making on contracted pricing and
other long-term decision making. The disadvantage is that the
holistic model does not take into account how the location values
will vary over time based on fluctuations in capacity and demand.
Both the holistic and dynamic models can be used not only for
calculating the marginal value of a unit of capacity, but they also
double as scenario testers to help answer a range of longer-term
tactical and strategic questions.
[0027] 3. Applications
[0028] FIG. 1 depicts a block diagram illustrating how the dynamic
and holistic models can be used in a number of example
applications. Dynamic network value engine (NVE) 102 calculates the
marginal value of a unit of capacity at a given location at a given
time, Val(i), by solving the dual linear program of the dynamic
model (described above). Dynamic NVE 102 thereby creates a matrix
containing the calculated marginal values Val(i) for a unit of
capacity at each location and each time increment. Dynamic NVE 102
calculates marginal values going into the future for a
predetermined number of days (for example, 14 days). Dynamic NVE
102 periodically updates the matrix values by resolving the dual
linear program. Dynamic value engine 102 thereby continually
updates the matrix in the background. By periodically updating the
matrix, Dynamic NVE 102 thereby updates the marginal values based
on changing market conditions, where trucks are being sent, changes
in demand and other network effects. Dynamic NVE 102 can be
performed by any computer or other processor capable of performing
the necessary computations with the required speed.
[0029] Load Value Calculator 104 calculates the profitability of
transporting a given load from A to B according to the equation
(described above): profitability=Revenue -Cost-Val (A)+Val (B).
Load Value Calculator 104 obtains the marginal values Val(A) and
Val(B) by pulling the appropriate number off of the matrix created
and updated by Dynamic NVE 102. Thus, Load Value Calculator 104
provides the capability to quickly and accurately determine the
profitability of a given load in real-time. The profitability
determination produced by Load Value Calculator 104 can then be
used by a number of useful applications, described below.
[0030] a. Sales Application 106
[0031] A carrier uses sales application 106 to sell large freight
commitments for the next year. Typically a carrier will send a
customer service representative (CSR) from the sales department to
visit a shipper to capture large freight commitments for the next
year. By using the profitability values determined by load value
calculator 104, the CSR can first determine which freight shipments
from which shippers are the most profitable. For example, the CSR
may determine that Retailer A has a shipment from San Francisco to
Denver that is highly profitable. This will prompt the CSR to make
a visit to Retailer A to discuss a shipment contract for that
particular lane.
[0032] The values calculated by load value calculator 104 also
allow the CSR to determine what prices to offer for each shipment
and what prices to accept. For example, the load value calculator
104 could provide to the CSR a range of prices and the potential
profitability of the shipment at each price. The CSR can thus
effectively use the data provided by load value calculator 104 to
offer prices and accept prices for shipment contracts for freight
for the next year at an optimal profit.
[0033] b. Solicit Application 108
[0034] Another application is solicit application 108. Solicit
application 108 is used by the carrier when the carrier has unused
capacity in a given region. For example, on a given day, a carrier
might have 10 trucks that are sitting idle in a particular
location. In this situation, the carrier will make calls to
shippers in that region that have freight to be shipped in order to
find freight for the idle trucks to ship.
[0035] In a given specified region, a carrier would like to
determine the optimal shippers to solicit based on 1) which
shippers are most likely to provide freight and 2) what is the
value of the freight that the shippers will provide on the lanes.
Solicit application 108 uses the values calculated by load value
calculator 104 to generate a list prioritizing the value of various
shipments from various shippers. A CSR can use this list to
determine which shippers to call, what prices to offer, and what
prices to accept. In other words, this list allows the CSR to
pursue the most profitable loads in order of priority.
[0036] As one example, suppose a carrier has an idle truck in San
Francisco. The CSR knows that shipper A has freight going from San
Francisco to Los Angeles, and they also have freight that's going
from San Francisco to Denver. The Load Value Calculator 104
calculates that the shipment from San Francisco to Los Angeles
would be much more profitable than the shipment from San Francisco
to Denver. The CSR could then use this information to solicit
shipper A to make an offer to ship the freight that is going from
San Francisco to Los Angeles.
[0037] c. Spot Price Application 110
[0038] In the spot price situation, a shipper calls a CSR at the
carrier because the shipper has a load that he needs moved, for
example, from Denver to Dallas. The shipper asks to find out the
spot price rate of the carrier. For example, the shipper has
freight that the shipper needs moved tomorrow, and the shipper
inquires about the spot price rate for shipping that freight. The
CSR at the carrier can then use spot price application 110 to
determine a spot price. Spot price application 110 uses Load Value
Calculator 104 to quickly determine a spot price rate which
produces a sufficient profitability for the carrier.
[0039] d. Order Accept Application 112
[0040] Most of the carrier's shipments will typically not be spot
pricing. Spot pricing usually involves shippers who have not
previously done business with the carrier. Generally the carrier
controls which freight it ships not through spot pricing, but
through contracted rates. The carrier mostly deals with shippers
using contracted rates. For example, on a particular day, shipper A
might have six loads that it needs to move from Denver to Los
Angeles. Shipper A has a contracted rate with carrier A. Shipper A
calls carrier A with a request to ship the six loads. The carrier
can decide whether or not it wants to take the loads at the
contracted rate. If the carrier has capacity available, it can
accept the loads. Carrier A, however, might decide that shipper B
has a shipment which is more profitable. Order accept application
112 uses Load Value Engine 104 to determine which shipments from
contracted shippers to accept Typically, order accept application
112 will accept orders from about one to fourteen days in the
future.
[0041] e. Mode Select Application 114:
[0042] Mode select application 114 is an application that selects a
"mode" for shipment of a particular load that the carrier has
agreed to transport. The carrier has several different "modes" by
which it can transport freight. Example modes include Solo, Team,
Canadian, Rail, Regional and 3rd Party. Solo mode is the standard
mode. Solo mode is a trucker with a truck and a trailer. Team mode
is a team consisting of two drivers in a single truck. Regulations
provide maximum time limits that drivers can drive without resting.
In team mode, the truck can be kept driving for extended periods of
time. The team is therefore more efficient, but is also more
expensive.
[0043] Rail mode is where the driver picks up a trailer with a
truck, shuttles the trailer to a rail yard, puts the trailer on a
train, and the train takes it across the country close to its
destination. The trailer is then picked up at the rail yard by
another truck and shuttled to its final destination.
[0044] Canadian mode is a truck with a Canadian driver. Under a
trade agreement between the U.S. and Canada, the regulations
require that a Canadian driver cannot pickup a load in the U.S. and
deliver that load within the U.S. Similarly, an American driver
cannot pickup a load in Canada and deliver the load in Canada.
However, a Canadian driver is allowed to pick up a load in the
United States and deliver the load in Canada. Therefore, if the
carrier has a load that's going, for example from Indianapolis to
Toronto, then the carrier must decide whether to have a Canadian
driver that's in the Indianapolis area take that load or whether to
put a U.S. driver on that load and then have the U.S driver take a
load out of Toronto back into the U.S.
[0045] Mode select application 114 uses load value calculator 104
to calculate the profitability of particular loads that the carrier
has agreed to transport for each available mode. Mode select
application 114 then selects the most profitable mode. Mode select
application 114 performs this selection for all loads that the
carrier has agreed to transport typically two to five days into the
future.
[0046] f. Load Assign Application 116
[0047] The Load Assign Application 116 looks at all the loads that
the carrier has committed to ship, typically over the next one or
two days. Load Assign Application 116 then selects a particular
driver and unit of capacity for each load. In other words, for a
given load from A to B in a given mode, Load Assign Application 116
selects an appropriate driver and unit of capacity to take the
load. When assigning a driver to the load, Load Assign Application
116 takes account of various factors such as eventually getting a
driver back to his or her home and using a big enough truck to
transport the load.
[0048] g. Load Track Application 118
[0049] Load Track application 118 tracks loads and the completion
of execution of loads. Load Track application 118 tracks all of the
data concerning the execution of the load. This data is then stored
and fed back to dynamic NVE 102. This allows dynamic NVE 102 to
constantly update its matrix of marginal values to update changing
market conditions and network effects.
[0050] 4. Order Process: Steps 120-132
[0051] The order process, comprised of steps 120-132, is the
process of using the various applications to receive offers, decide
which offers to accept, and assign modes and drivers to accepted
loads.
[0052] Order capture step 120 is the process of capturing orders
from either sales application 106, solicit application 108, or spot
price application 110. Order capture 120 is simply the receiving of
an offer to move freight moved from A to B for a specified price.
After the order has been captured, then in order accept step 122,
the decision is made whether or not to accept the offer. For
example, shipper A may call the carrier and ask for the spot price
to move a load in a particular lane. The carrier captures this
order in step 120. Later that day, the carrier calls shipper A back
and either accepts or rejects the offer in order accept step 122. A
big company may send a list to the carrier of all the shipments
they need shipped tomorrow. For example, it could list three
hundred shipments. The carrier will use load value calculator 104
to determine which offers to accept in order accept step 122.
[0053] In mode select step 124, mode select application 114 is used
to select the optimal mode for the loads that have been accepted by
the carrier. In mode assign step 126, the optimal mode is assigned.
For example, in mode assign step 126, a team can be assigned to
ship a load tomorrow from A to B. Load assign step 128 assigns the
actual drivers and trucks that will take particular loads. Note
that steps 124, 126, and 128 can all take place prior to order
acceptance in step 122. For example, the carrier might want to
determine the profitability of various modes before accepting an
order.
[0054] In execution step 130, the load is actually shipped. In load
completion step 132, the customer is billed for the shipment and
the tracking data from Load Track 118 is stored. As described
above, the tracking data is used by dynamic NVE 102 to update its
matrix to reflect changing network conditions. The data is also
used to update demand forecasts.
[0055] 5. Webcrawler Feature
[0056] A number of web sites on the World Wide Web currently
feature business-to-business exchanges for freight. These web sites
allow shippers to post loads that they need shipped. Carriers can
then review these loads and decide whether to accept any offers to
ship. Some of these exchanges allow the parties to make binding
commitments using the web site. Other exchanges merely provide a
phone number of the shipper or carrier, so that commitments can be
made offline. The web site might also allow a carrier to search for
loads meeting certain specifications. For example, the carrier
could request to view all of the freight being shipped out of
Chicago over the next week.
[0057] The system of the present invention illustrated in FIG. 1
can be used to implement a "webcrawler" feature. The webcrawler
scans a business-to-business exchange database for shipping
freight. The webcrawler uses load value calculator 104 to determine
which loads are profitable, and which are the best loads to accept,
if any. The loads that are highly profitable can immediately be
grabbed. If the carrier has its own business-to-business database,
the carrier can scan this database using the webcrawler.
[0058] The carrier could also use the webcrawler to scan other
business-to-business exchange databases connected to the Internet
(or any other network). A program which performs this type of
mining of information from databases on the Internet is sometimes
referred to as a "spider" or "bot."
[0059] 6. Conclusion
[0060] Although specific embodiments of the present invention have
been described, it will be understood by those of skill in the art
that there are other embodiments that are equivalent to the
described embodiments. Accordingly it is to be understood that the
invention is not to be limited by the specific illustrated
embodiments, but only be the scope of the appended claims.
* * * * *